What role should listening play in speaker design

[RexrothPigeon]

I think that’s likely quite exaggerated.

no, i'm very sure it's not, but the fact that we will never know if you're right or i'm right says it all

 

[MattHooper]

no, i'm very sure it's not,

Let’s say you and I are in the same room comparing two loudspeakers - a new Bowers & Wilkins 801 D4 Signature loudspeaker

Vs a Talon Khorus speaker

Do you actually think it’s implausible that after we’ve both listened to the speakers that I couldn’t give a description which captures to some degree the differences we are both able to perceive ?

If measurements like those don’t tell us anything about how we’re going to perceive it, what’s the point about caring how a speaker measures?

When I listened to the Talon speakers I perceived a notably down-tilted response emphasizing the upper bass/low mids with an obvious roll off in the highs (and I could describe the effect that had with various music tracks). When I listened to the B&W I noted an obvious rising trend in the highs with peakiness in that region. (among other differences between the speakers) . And that had certain effects on the sound of music through the speaker.

Would you would hear things just the opposite?

Is there any reason to have such an expectation? Just how different do you think we would hear these speakers?

(I’ve heard both of these speakers before I saw the measurements and both of them sounded like they measure).

Or what about Erin of Erin’s Audio Corner.
His subjective impressions and descriptions before he measures loudspeakers seem to track the nature of the measurements quite well.

Do you think that after you listened in the same room with Erin to Revel PerformaBe F226Be Speakers

And then to Zu Audio Method Speakers

(you can find measurements in those links)

Do you think that Erin’s descriptions of the subjective sonic differences between those speakers would not track what you hear at all?

At this point, you may want to say “ but if you’re showing me the measurements that’s all I need” that of course isn’t the point.
The point is whether you truly think it’s implausible that somebody - me, Erin, Amir or whomever - could describe what we hear in a way that would match to some degree characteristics you’d be perceiving as well.

Again, if the measurements don’t predict to some degree what you and I will perceive… then why care about measurements at all?

 

[Arindal]

Getting this wrong and then relying on off-axis to compensate is a completely crapshoot.

Agreed, and a compensation as you describe it, was never suggested as useful. We are rather talking about both including imbalances or errors, and speaker designer has to find a compromise between them.

On-axis neutrality is the most important characteristic of a speaker.

Let us agree on, it is the easiest one to identify subtle changes in, the best when it comes to finding correlations between listening and measuring, and the first one to notice when listening to unknown material on a new speaker. If you sub summarize that under ´most important´, may it be.

But: it is also the easiest to manipulate and correct. Buying hi-fi with built-in DSP, or even a software-based solution, is cheap and easy nowadays, and mandatory for the bass bands with the speakers in a room, anyways. You mentioned a -2dB dip @3K of a TAD speaker in your eyes being ´unacceptable´, if I understood you correctly. Well, looking at the graph, it would take me like 10 seconds to model PEQ filters to correct for that sufficiently.

So why bother about anechoic response at all (If we are not talking about sharp, pronounced deviations, most likely not to be correctable)?

No way can you predict the reflections in anyone's room. Today, many homes have open floor plans, meaning some of the reflections don't even exist!

Certainly true. But the inability to precisely predict level and tonality of the first reflections, does not mean these are negligible, or the reflections do not exist. They do, and having the averaged level of all windows proportional to the anechoic response, is an ideal starting point of getting to a balanced indirect soundfield as well (proper ratio of direct vs indirect is the second). This is where problems with existing speakers and rooms start.

This is why I keeps saying flat on axis and well behaved off-axis.

´Well-behaved´ is not defined. We could agree on a linear response in all the aforementioned windows hence constant directivity.

So, the moment ´well-behaved´ is implying decisive compromises, things get difficult. As wavelengths are decreasing with higher frequencies, companies tend to accept a continuously narrowing dispersion pattern over the audible band, as the resulting tilted lines look tidy and ´well-behaved´ on the graph. Which in my opinion results in the much disadvantageous reverb tonality imaginable, a tilted one with overrepresented lower midrange and severe lack of brilliance/treble.

f flat on-axis response makes off axis worse, then that is designer's job to fix. Not make on-axis worse as a cure.

That is basically impossible, and would translate to a perfect pointsource with perfect constant directivity, which is in combination pretty difficult, if not impossible, to design (for the obvious reason of diaphragms being much bigger than 0, and the variety of audible wavelengths is pretty vast).

Or could you name a solution for example of bring a speaker, delivering an overshoot of (lower) midrange compared to brilliance and treble bands due to increasing directivity index, to constant directivity, without compromising on the on-axis response?

 

[amirm]

That is basically impossible, and would translate to a perfect pointsource with perfect constant directivity, which is in combination pretty difficult, if not impossible, to design (for the obvious reason of diaphragms being much bigger than 0, and the variety of audible wavelengths is pretty vast).

I am asking for perfect on-axis, not off-axis. If you can't manage this, then better get out of the kitchen:

 

[amirm]

Or could you name a solution for example of bring a speaker, delivering an overshoot of (lower) midrange compared to brilliance and treble bands due to increasing directivity index, to constant directivity, without compromising on the on-axis response?

I am not advocating constant directivity. Sloping DI is fine with me. It would result in a sloping down overall response which is what is preferred anyway.

 

[amirm]

Let us agree on, it is the easiest one to identify subtle changes in, the best when it comes to finding correlations between listening and measuring, and the first one to notice when listening to unknown material on a new speaker. If you sub summarize that under ´most important´, may it be.

I don't mean this as a personal comment, but are you using machine translation? I can't follow half of what you write. "sub summarize?" "most important, may it be?"

 

[RexrothPigeon]

At this point, you may want to say “ but if you’re showing me the measurements that’s all I need” that of course isn’t the point.

No, that is exactly the point. Sharing subjective impressions is nothing more than a mishmash of adjectives whose efficacy hinges partly on how invested we both are in a shared shorthand lexicon of metaphors ("warm," "clinical," etc.), the old "dancing about architecture" cliche. I just really don't want to learn that stuff or be influenced by it, since it's the stock-in-trade of the grifters and snake oil salesmen and I don't have any need of it anyway.

 

[SigbergAudio]

No, that is exactly the point. Sharing subjective impressions is nothing more than a mishmash of adjectives whose efficacy hinges partly on how invested we both are in a shared shorthand lexicon of metaphors ("warm," "clinical," etc.), the old "dancing about architecture" cliche. I just really don't want to learn that stuff or be influenced by it, since it's the stock-in-trade of the grifters and snake oil salesmen and I don't have any need of it anyway.

I don't think the point is whether you find it useful to hear subjective remarks on sound. The point is rather that the fact that two people are able to share somewhat meaningful information about the sound independently in a sighted listening, implies that what we are able to hear in a sighted listening isn't 100% random.

 

[SigbergAudio]

Having watched this thread in the last day or so, it appears that the HiFi audio subjectivists are alive and well.

They all came out of hiding from the woodworks cheering on feel, gut, artistic expression and emotions as the anthem of speaker design.

It is interesting how differently we can read a thread (is it the famous bias that plays out even when reading, perhaps?).

The ones I've identified so far in this thread (maybe two?), haven't really been in the woodworks, they've been open about their opinions in many other threads on this forum, nothing new at all.

We have gone back few decades of scientific progress with this thread. I bet the next thing to follow is Danny Ritchie to get promoted as the new standard bearer of speaker design.

Overall I can't really recollect reading many posts (if any at all) in this thread that warrants such a claim. I haven't seen anyone disregarding the importance of measurements or anything "scientific" in this thread. Care to share a couple of examples of what you are reacting to?

I would say the complete opposite, this thread has helped provide a bit more nuance on topics that are not so black and white as some suggest. It seems we sometimes forget that one of the cornerstones of science is to be curious. It seems we also sometimes forget that the point of all of this is a leisure activity.

 

[RexrothPigeon]

The point is rather that the fact that two people are able to share somewhat meaningful information about the sound independently in a sighted listening, implies that what we are able to hear in a sighted listening isn't 100% random.

then the point is arguing against a straw man because "100% random" is not an argument anyone is making here

look, when you have an option that is clear and accurate and not random, and you have another option that is ambiguous and metaphorical and [some frequency] random and [some amount] distorted by inescapable biases, then you take the first option and ignore the second since it will only diminish reliability. that's the situation we find ourselves in today in audio.

 

[Vuki]

It seems to me that those advocating speaker design w/o listening haven't designed any speaker ever. I could be wrong?

 

[SigbergAudio]

then the point is arguing against a straw man because "100% random" is not an argument anyone is making here

look, when you have an option that is clear and accurate and not random, and you have another option that is ambiguous and metaphorical and [some frequency] random and [some amount] distorted by inescapable biases, then you take the first option and ignore the second since it will only diminish reliability. that's the situation we find ourselves in today in audio.

It's not a very large straw man, because some strongly suggests this to be the case.

The entire context of this discussion is loudspeaker development. So if you were a loudspeaker designer (assuming you're not, I don't know), you would develop the speaker exclusively through measurements, presumably working your way towards the best possible on-axis response? And then call it a day?

Say you found yourself in a situation where you had conflicting compromises in the measurements to choose from. Would you select one of them at random unless a panel of listeners conducting double blind tests between the two were available to you? Would you conclude that "I cannot trust my ears, so there is no point of me listening to the speakers to consider which alternative sounds the best." ?

 

[Newman]

no, i'm very sure it's not, but the fact that we will never know if you're right or i'm right says it all

Did you notice how the reply you got was a long-winded anecdote that 'sounds logical', laced with hypothetical whataboutisms, but zero valid evidence? Not good enough (but highly characteristic of the denialist's tactical armoury).

Your statements were consistent with the science as known from valid experiments. The sighted listening effect (SLE) is stronger than most people are likely to accept without it being proven to them personally via experimental participation. Hence the never-ending resistance online...because so few people will have that opportunity to experience it personally in a validly constructed experiment.

The confident likelihood that you can construct listening experiments with differences so clear that many people will correctly pick them in a sighted listening test, only brings the science of the SLE into dispute in the minds of those who wish it wasn't as strong as it is. It is an invalid argument and not evidence that the SLE is weak, or overstated.

I don't think the point is whether you find it useful to hear subjective remarks on sound. The point is rather that the fact that two people are able to share somewhat meaningful information about the sound independently in a sighted listening, implies that what we are able to hear in a sighted listening isn't 100% random.

Nobody said it is 100% random. You can't be twisting the argument so maliciously by mistake, or simply because you fell for Matt's debating tactics. In fact it is 100% non-random, in that one's cognitive biases will always work in the same directions at any one point in time (they will often evolve over time). There will be a lot of variations between individuals, but then again, there are often common threads among cognitive biases that result in many people being influenced in the same way by the same non-sonic variables. That's not '100% random'.

For example, take the unvalidated anecdote that was mistakenly tossed up as evidence of....something. Let's say one of those speakers was the personification of widely-respected audiophile credibility, and the other was a cheap and cheap-looking, obviously no-name Chinese, much smaller speaker with visual aspects that make one question the sense of its design. If the Chinese speaker had the rising response, it would be unsurprising if it were described in sighted listening comparisons as relatively tinny, sharp, and anaemic in the bass, whereas the audiophile speaker might be described as solid and smooth with high liquidity and musicality. But if the Chinese speaker had the falling response, ie the sonics were reversed, don't be surprised if it is described as relatively muffled and boomy with an absence of detail, whereas the audiophile one suddenly has astonishing clarity and transparency with a lively, foot-tapping bass.

That is the order of magnitude of SLE strength that we need to account for. To pretend that we can be immune to it is wishful thinking. Maybe even denialism.

cheers

 

[Geert]

Sharing subjective impressions is nothing more than a mishmash of adjectives whose efficacy hinges partly on how invested we both are in a shared shorthand lexicon of metaphors

It doesn't have to be like that. Look at the example of Erin’s Audio Corner reviews Matt gave as an example. It’s perfectly possible to describe the perceived frequency response of a speaker using universal names of frequency ranges (sub, low, low-mid, mid ...) or even frequencies. When audiophiles revert to fluffy metaphors, they only show they're not as experienced as they think they are.

 

[TimVG]

´Well-behaved´ is not defined. We could agree on a linear response in all the aforementioned windows hence constant directivity.

The largest study covering this is "A Multiple Regression Model for Predicting Loudspeaker Preference Using Objective Measurements" by Dr. Sean Olive.
If you analyse the study the preferred slopes can be deducted. Conveniently in Vituixcad the preferred slopes can now be set standard as "shadow slopes" so speaker data can be referenced against it.

Example of the Ascilab C8C overlayed: excellent agreement overall.

Neumann KH150: excellent agreement in nearfield, however too little energy from lower treble upwards => good for nearfield, probably a little dark sounding for most in the farfield.

This response is meant as information, not discuss the merits of the study. If anyone wants to do that, take it up with Dr. Sean Olive.

 

[RexrothPigeon]

It’s perfectly possible to describe the perceived frequency response of a speaker using universal names of frequency ranges (sub, low, low-mid, mid ...) or even frequencies.

Yes, the more closely subjective description resembles information that is better conveyed in graphic or numerical form, the more accurate it is (thus underscoring its pointlessness).

 

[Mnyb]

It's not a very large straw man, because some strongly suggests this to be the case.

The entire context of this discussion is loudspeaker development. So if you were a loudspeaker designer (assuming you're not, I don't know), you would develop the speaker exclusively through measurements, presumably working your way towards the best possible on-axis response? And then call it a day?

Say you found yourself in a situation where you had conflicting compromises in the measurements to choose from. Would you select one of them at random unless a panel of listeners conducting double blind tests between the two were available to you? Would you conclude that "I cannot trust my ears, so there is no point of me listening to the speakers to consider which alternative sounds the best." ?

How personal is your process ? You seems to be a one man operation for the moment .

Is it about what you prefer or what your customers prefer . Or what in some larger context is correct ( some posters here seems alluding to that this exist , but never defined it . Some say vaguely Tool and Olive some has an undisclosed personal standard like @Arindal ).

If you managed to crow your company ( I whish you well products seems amazing btw ) you better get that listening panel .

I suppose you actually tries to get second opinions .

Btw I 100% believes in scientific methods but as an amateur I want to have definitions rehashed now and then otherwise I’m a bit lost.

So how should a speaker behave . I have Toole’s book but not much else as references.
Another point about circle of confusion will this also eventually change speaker design. If it understood what we prefer to listen to with already made recordings when studious also get better speakers and listening room that translates to even better sounding recordings.
Do anyone foresee any large shift in speaker preferences or are we in some kind of iterative status ? Like how wide dispersion and how much should directivity slope .

 

[Geert]

Yes, the more closely subjective description resembles information that is better conveyed in graphic or numerical form, the more accurate it is (thus underscoring its pointlessness).

It’s not pointless when measurements (incl. directivity) aren’t available, and it’s also interesting to see how perceived sound relates to measurements (since a lot of audiophiles believe that relationship is very poor).

 

[Mnyb]

The largest study covering this is "A Multiple Regression Model for Predicting Loudspeaker Preference Using Objective Measurements" by Dr. Sean Olive.
If you analyse the study the preferred slopes can be deducted. Conveniently in Vituixcad the preferred slopes can now be set standard as "shadow slopes" so speaker data can be referenced against it.

View attachment 521051

Example of the Ascilab C8C overlayed: excellent agreement overall.

View attachment 521052

Neumann KH150: excellent agreement in nearfield, however too little energy from lower treble upwards => good for nearfield, probably a little dark sounding for most in the farfield.

View attachment 521053


This response is meant as information, not discuss the merits of the study. If anyone wants to do that, take it up with Dr. Sean Olive.

Thankyou this helps me .

Now if speaker companies could flat out tell us the listening distance, pro audio seems to optimise home audio won’t tell ( use our speaker everywhere )

 

[jean-benoit]

i dont think its a shame, that andrew doesnt design his speakers with a science first approach.
after all its a lifestyle product, not meant for professional use.

although i love seeing brands like KEF going in a different direction. im not mad at a brands like mofi or buchardt for choosing subjective impressions over measurements. even if that doesnt make sense. both companys have proven, that they still are capable of producing speakers that can measure good.
dont forget – they are lifestyle products, just like a sofa or a tv.

i also wouldnt compare them to a brand like neumann, because they do not compete in the same area.

personally, i liked the look of the mofi sourcepoint 8 a lot. however, this new speaker doesnt look good to me and i wouldnt put it in my living room just for that.